Comment by fpgaminer | Hacker News Reader

Comment by fpgaminer | Hacker News Reader

Parent Full thread

I recently looked at the landscape of smart door locks, and came to the same conclusion as @Animats did in his comment. Actually, I would rephrase it to say _all_ smart door locks suck. All I really want right now is an RFID/NFC deadbolt, purely for convenience. Seems simple enough, and yet nothing secure exists.

So I'm now set on trying to build my own lock. I'm thinking of just getting a really nice, mechanical deadbolt, and then building something similar in design to Lockitron, where it just sits on top of the deadbolt, augmenting it with digital authentication. For the keycards I plan to combine a low power MCU with NFC and Ed25519 cryptography for authentication. I found a variation of the MSP430 chips with built-in NFC capability which are designed to run battery-less (they harvest energy from the NFC signal). My rough calculations suggest that they'll be able to soak up enough power to do an Ed25519 signature in 0.25 seconds; fast enough for me. I'll just have the lock send a random challenge to the keycard, it responds with a signature of the challenge, and then the lock can verify the signature and confirm that the card is legit. It should be simple enough to mold the MSP430 and supporting electronics into a thin 1" diameter disc which I can hook to my keychain or put in my wallet. Awesome! The lock itself is just another MCU with NFC capabilities, battery, and motor.

That setup satisfies my desire for a convenient, digital lock, while maintaining real security (Ed25519 with real random challenges, versus existing RFID solutions which use 48-bit keys, 3DES, suffer replay attacks, etc). I'm surprised these things don't already exist. I only had to spend an hour to find the MSP430 chip variation and sketch out this system...

We previously sold Lockitron units in the form you describe as a kit on Adafruit and Sparkfun. Send me an email I'll get you set up.

As for finding a secure NFC lock - I agree, it's slim pickings. The industry moved toward bluetooth in part because it lets you do random challenges defined by (your) software rather than (their) hardware.

What benefit do you hope to get from your smart lock? I stayed at an Airbnb with a simple electronic combo lock.

Plus: no key to pass me Plus: can change code for each new guest

It's not a smartlock tho but maybe I don't understand what a smartlock is. A friend as an internet connected lock

Plus: can open remotely (package delivery guy calls. From phone friend remotely let's him into garage Plus: can set time based codes (code only works on a particular day/time range)

Minus: it's on the internet and controlled through a third party service

What will yours do? I'd love one that's online somehow but secure

I'm only going to start with electronic keycard access. For me, being able to replace big, sharp keys with small, smooth cards/discs of plastic is a big win. No fiddling with my keychain, just walk up to the door, press my wallet/keychain/pocket to the door and tap. I want that convenience, without having to sacrifice security (which is the case with all such consumer locks I found).

Calling that a smart lock is a bit of a stretch. Keycard access is old school, but was only used in commercial settings until recently. It's just that the consumer market has dubbed anything that isn't a plain mechanical lock a "smart" lock when applied to people's homes.

Internet connectivity would be something I add later. I'm not as motivated to put the device online, as the benefits as you listed them are a bit fringe for me. Assuming I complete this project it will certainly get open sourced, so perhaps other users will motivate the addition of such features.

I'm not really sure why you need an electronic lock on an airbnb, you're either there to flip the place or you have an onsite person to flip the place, well in large numbers of airbnbs. The ones I can think of that don't need it wouldn't have internet generally. Interested in what you see as the benefit over one of those punch code locks that hold the keys that are like $14 on amazon or a schlage deadbolt for $150.

The benefit to a code lock for me over a box with a code that holds the keys is there's no key to lose nor any key to forget to return. I don't even have to carry the key when I go out. One less (three less?) things to worry about. I actually locked myself out of the last AirBnb I was at. I forgot the key and while that meant I didn't lock the door to the apartment there was also an automated door to the building. Going out it auto-opens, going in it requires a key. I just sat in front of the building for 30 minutes until someone else opened the door.

I've stayed at about 15 AirBnBs, most of them I didn't have to meet the owner. But, also most of them used a lock box with a code instead of just a door code lock.

Looking at the statistics, are a significant number of break in's happening via lock hacking or via the good old fashioned kicking in the front door in your neighborhood? We have had break in's in our neighborhood and lots of conversations with the police. The former is unheard of and the later is the common practice. One thing that really surprised me was that you would think that burglars would go through a window if a door is difficult to penetrate, but once again in my neighborhood that is a rare occurrence even with a large number of homes with easy to get at windows. My point is that this is not likely to be an attack vector unless the other mechanisms are more difficult at the majority of the houses in your neighborhood, or you truly have significantly more expensive items than your neighbors (and it is obvious that you have these items).

This. So much this.

Fancy technology defeated with a rock, or more likely a crowbar.

Also keep in mind, the vast majority of break ins are over in 5 minutes or so. It's literally just grab anything you can easily carry that's valuable and easy to see. So your big screen television is safe. Your laptop on the coffee table isn't. The laptop in the coffee table? Shrug.

Using public key crypto for this seems like overkill. If you're willing to use symmetric crypto, a DESFire EV1 card is probably the way to go, and you can get those off the shelf for a dollar or two. Or, if you want public-key signatures, Yubikey devices can do it.

I'm curious, though: why a gadget that fits on the knob? You can get off-the-shelf electric strikes and off-the-shelf electric fail-secure locks in AC and DC variants quite easily.

Public key crypto is exactly what should be used in this application.

Specifically, symmetric crypto would require exchange of secret material between the keycard and the lock at least once. Whether the lock is telling the keycard what secret to use, or the keycard tells the lock what secret it's using. Both scenarios introduce vulnerabilities if someone is able to listen in on that exchange, and one of those scenarios means the keying material in the keycard needs to be dynamic, which introduces problems.

Public key crypto on the other hand is dead simple, hard to get wrong, and a perfect analogy for this application. To enable a keycard you just have the lock read its public key. To auth, you just verify a signature. No secret material is ever exchanged which means no MITM and the keycards remain static devices which means it's easier to make them unclonable.

RE: DESFire EV1. I don't trust MIFARE. Old habits die hard, and MIFARE has very bad habits. Plus all the above.

> No secret material is ever exchanged which means no MITM

That's a bit of a stretch. A MITM could replace the public key that gets transferred. Admittedly, making it look like enrollment worked is awkward, but there is certainly no cryptographic prevention of a MITM.

I should also point out that, even if you choose to use public key crypto, using digital signatures in particular is not necessarily the right choice. The property you want is authentication, not necessarily a non-repudiable signature. For example, the private key holder could authenticate by decrypting an encrypted challenge or completing a DH exchange.

If you're using Curve25519, doing the DH operation is a bit cheaper and less fiddly than running Ed25519, which is a nice benefit. If you're using P-256, it's a much bigger deal: ECDH on P-256 is much safer than using ECDSA, especially on a limited device, as an ECDSA signature using a poor RNG can negate the security of the entire system.

Excuse my ignorance but what will you do if the battery fails or your firmware has a bug and the deadbolt won't open? smash the door down?

You could always try unlocking the door with your key first.

which means do i need tocarry both Key and whatever fancy gizmo unlocks the door?

There are workarounds (e.g. the key is also the dongle, like with many cars) but for the sake of this argument, yes, you'd need to carry both.

The car analogy is a good one, actually, because keyless entry has become commonplace in that arena. It's less secure, but so darned convenient. These days, it has become nearly impossible to get a mid-tier, or better, automobile without keyless entry; though it won't happen as quickly, we can expect the same trend with our home door locks. After that, we'll be replacing our locks every 4-6 years, or at around 150,000 miles. Yey, more stuff to go obsolete and throw away!

Generally speaking, the privacy and security concerns that industry continues to ignore are turning me into a Luddite, so I won't be in line for an off-the-shelf IoT device anytime soon, if ever.

How secure is the mechanism to tapping or smashing? Does it fail open, or closed?

The whole unit is just a motor that actuates an existing deadbolt's thumbturn on the opposite side of the door (i.e. not accessible from the outside world). So failing open/closed is a question for the existing deadbolt, not this device.

Reply on news.ycombinator.com